Skip to main content
. 2021 Mar 22;10:e61590. doi: 10.7554/eLife.61590

Figure 3. Spatial proteomics and network covariation analysis reveal significant disruptions to the WASH complex and an endosomal module in SWIPP1019R mutant mouse brain.

(A) Tandem-mass-tag (TMT) spatial proteomics experimental design. Seven subcellular fractions were prepared from one WT and one MUT mouse (10mo). These samples, as well as two pooled quality control (QC) samples, were labeled with unique TMT tags and concatenated for simultaneous 16-plex LC–MS/MS analysis. This experiment was repeated three times (three WT and three MUT brains total). To detect network-level changes, proteins were clustered into modules, and linear mixed models (LMMs) were used to identify differences in module abundance between WT and MUT conditions. The network shows an overview of the spatial proteomics graph in which 49 different modules are indicated by colored nodes. (B) Protein module 38 (M38) contains subunits of the WASH, CCC, Retriever, and CORVET/HOPS complexes. Node size denotes its weighted degree centrality (~importance in module); purple node color indicates proteins with altered abundance in MUT brain relative to WT; red, yellow, orange, and green node borders highlight protein components of the CCC, Retriever, CORVET/HOPS, and WASH complexes obtained from the CORUM database; dashed black edges indicate experimentally determined protein-protein interactions; and gray-red edges denote the relative strength of protein covariation within a module (gray=weak, dark red=strong). (C) M38 displays decreased overall abundance in MUT brain. The aligned profiles of all M38 proteins are plotted together after sum normalization, and rescaling such that the maximum intensity is 1. Each solid line represents a single protein, measured in WT (teal) and MUT (purple) conditions. The estimated WT and MUT means are displayed in dashed teal and purple lines, respectively (WT-MUT Contrast log2Fold-Change=−0.12, T=−11.14, DF=2324, p-adjust=2.078×10−26; n=3 independent experiments). (D) Protein profile of WASHC4 (aka SWIP) plotted as relative (sum-normalized) protein intensity, rescaled to be in the range of 0–1 (WT-MUT Contrast log2Fold-Change=−1.517, DF=26, p-adjust=1.72×10−16; n=3 independent experiments). WT levels are depicted in teal, and MUT levels are depicted in purple. Shaded error bar represents the min-to-max values of all three experimental replicates. Significant differences in individual BioFraction WASHC4 levels are indicated with stars. ***p<0.001, MSstatsTMT p-value for intra-BioFraction comparisons with FDR correction.

Figure 3.

Figure 3—figure supplement 1. Western blot analysis of LAMP1 abundance across subcellular brain fractions.

Figure 3—figure supplement 1.

(A) Line graph depicting LAMP1 signal abundance normalized to total protein across subcellular biological fractions. LAMP1 pattern reveals no difference between MUT and WT in LAMP1 pelleting, but a significant difference in the abundance of LAMP1 in BioFraction 8 (two-way ANOVA genotype effect, F4,44=7.68, p<0.0001; Sidak’s multiple comparisons test, p=0.0006). Shaded region highlights subcellular fractions used in 16-plex TMT mass spectrometry experiments. Fractions 4–10 from one WT and one MUT mouse were analyzed for each TMT experiment. Data reported as mean ± SEM, error bars are SEM. (B) Western blots depicting LAMP1 abundance from three separate WT animals (Experiments 1–3). (C) Western blots depicting LAMP1 abundance from three separate MUT animals (Experiments 1–3).
Figure 3—figure supplement 2. Western blot analysis of EEA1 abundance across subcellular brain fractions.

Figure 3—figure supplement 2.

(A) Line graph depicting EEA1 signal abundance normalized to total protein across subcellular biological fractions. EEA1 pattern reveals no difference between MUT and WT in EEA1 pelleting (two-way ANOVA genotype effect, F1,4=0.290, p=0.6184). Shaded region highlights subcellular fractions used in 16-plex TMT mass spectrometry experiments. Fractions 4–10 from one WT and one MUT mouse were analyzed for each TMT experiment. Data reported as mean ± SEM, error bars are SEM. (B) Western blots depicting EEA1 abundance from three separate WT animals (Experiments 1–3). (C) Western blots depicting EEA1 abundance from three separate MUT animals (Experiments 1–3).
Figure 3—figure supplement 3. Western blot analysis of total protein abundance across subcellular brain fractions.

Figure 3—figure supplement 3.

(A) Line graph depicting total protein abundance across subcellular biological fractions. There is no difference between MUT and WT samples (two-way ANOVA genotype effect, F1,4=0.0094, p=0.9274), measured by mean fluorescence units (M.F.U.) per lane. Shaded region highlights subcellular fractions used in 16-plex TMT mass spectrometry experiments. Fractions 4–10 from one WT and one MUT mouse were analyzed for each TMT experiment. Data reported as mean ± SEM, error bars are SEM. (B) Western blots of total protein abundance across subcellular fractions from three separate WT animals (Experiments 1–3). (C) Western blots of total protein abundance across subcellular fractions from three separate MUT animals (Experiments 1–3). Total protein abundance per lane was used for normalizing endosomal and lysosomal markers.
Figure 3—figure supplement 4. Spatial proteomics experimental design and data analysis.

Figure 3—figure supplement 4.

(A) Experimental design used for tandem-mass-tag (TMT) spatial proteomics. Each experiment was designated as a mixture (Mix 1–3), with seven BioFractions prepared from one WT and one MUT animal (TMT channels: C1–7=WT; C9–15=MUT). The g-force pelleting speed used to obtain each BioFraction is indicated in the corresponding square (i.e. 5K=5000xg). Two sample-pooled quality control samples were used for normalization between experiments in channels 8 and 16 (QC1 and QC2). (B) Boxplot of protein intensity for all proteins in each sample showing no large differences within or between batches after normalization. (C) Principal component analysis of each sample reveals separation of all 7 BioFractions. (D) The variance attributable to the experiments’ major covariates for every protein. After normalization, the major source of variation for most proteins is BioFraction. After normalization, the variation attributable to differences between mixtures is negligible. (E) Principal component analysis of proteins showing the 49 modules identified by clustering the spatial proteomics network. (F) Matrix of comparisons used by MSstatsTMT to assess intra-BioFraction and overall protein differences between Control and Mutant conditions.
Figure 3—figure supplement 5. Analysis of spatial brain proteome reveals conservation of organellar compartments found in LOPIT-DC dataset.

Figure 3—figure supplement 5.

(A) Module 4 is 5.6-fold enriched for plasma membrane proteins (p-adjust=1.39×10−38) and displays no significant change in module-level intensity between WT and MUT conditions (p-adjust=0.323). (B) Module 5 is 2.4-fold enriched for nuclear proteins (p-adjust=3.48×10−22) and displays no significant change in module-level intensity between WT and MUT conditions (p-adjust=0.270). (C) Module 1 is 6.7-fold enriched for cytosolic proteins (p-adjust=1.06×10−89) and displays a slight change in module-level intensity between WT and MUT conditions (p-adjust=6.5×10−20). (D) Module 13 is 6.1-fold enriched for ribosomal proteins (p-adjust=2.0×10−10) and displays no significant change in module-level intensity between WT and MUT conditions (p-adjust=1). (E) Module 7 is 4.7-fold enriched for endoplasmic reticulum proteins (p-adjust=8.43×10−16) and displays no significant change in module-level intensity between WT and MUT conditions (p-adjust=1). (F) Module 40 is 6.8-fold enriched for mitochondrial proteins (p-adjust=5.40×10−15) and displays a slight change in module-level intensity between WT and MUT conditions (p-adjust=8.18×10−6). (G) Module 9 is 18.5-fold enriched for peroxisomal proteins (p-adjust=1.39×10−20) and displays no significant change in module-level intensity between WT and MUT conditions (p-adjust=0.474). (H) Module 8 is 27.6-fold enriched for proteasomal proteins (p-adjust=7.91×10−41) and displays no significant change in module-level intensity between WT and MUT conditions (p-adjust=1). For all modules, light purple nodes represent proteins with decreased abundance in MUT samples, dark purple nodes represent proteins with increased abundance in MUT samples, blue node borders highlight proteins with sample organelle designation as (Geladaki et al., 2019), experimentally determined protein–protein interactions (HitPredict), and gray-red edges denote the relative strength of protein covariation within a module (gray=weak, dark red=strong). Corresponding profile plots for each module depict individual scaled protein intensities in light teal (WT) and purple (MUT) lines, with the estimated value of WT and MUT conditions shown as dashed lines.